Tweeting about alcohol: Exploring differences in Twitter sentiment during the onset of the COVID-19 pandemic.

Abstract

This study explores pandemic-related changes in Twitter communication by examining differences in emotional, psychological and social sentiment between alcohol-related tweets and a random sample of non-alcohol tweets during the onset of the COVID-19 pandemic. Two equivalent size sets of English-language, COVID-specific tweets posted between February 1st and April 20th, 2020 are examined. The first set includes 1.5 million tweets containing alcohol-related keywords, while the second set does not contain such references. LIWC software analyzed the tweets for sentiment factors. ANCOVAs were used to determine whether language use significantly differed between the sets, considering differences in the pandemic period (before or after the pandemic declaration) while controlling for the number of tweets. The study found that tweets in the 40 days after March 11, 2020 contained more authentic language, more affiliation-oriented language, and exhibited more positive emotion than tweets in the 40 days pre-declaration. Alcohol-related status was a significant factor only when tweets contained personal concerns, regardless of pandemic period. Authenticity levels increased significantly in alcohol-related tweets post-declaration. The findings suggest alcohol may play a lesser role in the expression of psychological, social, and emotional sentiment than the pandemic period, but interaction between authentic language and alcohol references may reflect an increased use of alcohol for coping.